JP2966711B2 - Image contour extraction method and apparatus - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image outline extracting method for extracting an outline of an object using image data consisting of coordinate data and luminance data of the surface of the object input by a three-dimensional image input device. Regarding the apparatus, for example, when the object is a mechanical part or the like, image contour extraction used to achieve the purpose of creating illustration data for insertion into a book such as a manual of a mechanical apparatus using the part A method and an apparatus therefor.

[0002]

2. Description of the Related Art When contours are extracted from image data consisting of coordinate data and luminance data on the surface of an object, generally, gradients of the coordinate data and luminance data are calculated and derived for each pixel, and the gradients are calculated. In order to extract a steep part as a contour, a threshold is set and binarized. That is,
Pixels having inclination data larger than the threshold value are extracted as contours. Here, conventionally, an average value of gradients of all pixels is obtained, and a value larger or smaller than the average value is set as a threshold value, or a value such that data is divided into two by taking a histogram of inclination data Was set as the threshold.

[0003]

However, since the luminance data includes a value to which the shape represented by the coordinate data contributes (the luminance may be different due to the influence due to the shape), According to the prior art described above, the pattern drawn on the surface of the object (in addition to characters and figures drawn in a plane,
It has been difficult to set a threshold value for accurately extracting only characters (including characters and figures represented by irregularities). For example, when extracting the contour of a pattern drawn on the surface of an object, it is necessary to set a threshold value in accordance with the characteristics of the drawn pattern. This is because no appropriate threshold is set. SUMMARY OF THE INVENTION An object of the present invention is to provide an image contour extraction method and an apparatus for accurately extracting a contour from image data composed of coordinate data and luminance data on the surface of an object, while solving the above-mentioned conventional disadvantages.

[0004]

In order to achieve this object, a feature of an image contour extracting method according to the present invention is that image data comprising coordinate data and luminance data of a surface of an object inputted by a three-dimensional image input device is provided. A first step of calculating and deriving a gradient of coordinate data for each pixel, binarizing the gradient with a first threshold to extract shape contour data, and a step of extracting luminance data for each pixel. A second step of calculating and deriving a gradient, masking a portion corresponding to the contour of the shape extracted in the first step from the gradient, and then binarizing the pattern with a second threshold to extract pattern contour data And a third step of extracting the contour data of the object by synthesizing the contour data of the shape extracted in the first step and the contour data of the pattern extracted in the second step. is there.

Further, the characteristic configuration of the image contour extracting apparatus according to the present invention is that the image data consisting of the coordinate data and the luminance data of the surface of the object inputted by the three-dimensional image input apparatus is used for each pixel. A first processing circuit for calculating and deriving a gradient, binarizing the gradient with a first threshold to extract contour data of the shape, and calculating and deriving a gradient of luminance data for each pixel, A second processing circuit for masking a portion corresponding to the contour of the shape extracted by the one processing circuit and then binarizing the pattern with a second threshold to extract the contour data of the pattern; and a shape extracted by the first processing circuit. And a third processing circuit for extracting the contour data of the object by synthesizing the contour data of the pattern and the contour data of the pattern extracted by the second processing circuit.

[0006]

The coordinate data of the surface of the object includes only information on the shape. Therefore, in the first stage (first processing circuit), the gradient of coordinate data is computed and derived for each pixel, and the gradient is binarized by the first threshold, whereby the contour data of the shape can be accurately extracted. . On the other hand, the brightness data on the surface of the object includes those based on the shape and those based on the pattern or the like. Therefore, in the second stage (second processing circuit), a gradient of the luminance data is calculated and derived for each pixel, and the gradient corresponding to the contour of the shape extracted in the first stage (first processing circuit) is calculated from the gradient. By masking the parts that do, the influence of the shape data can be removed from the luminance data,
If the gradient after removing the influence of the shape data is binarized with the second threshold value, the contour data of the pattern can be accurately extracted.
Therefore, in the third stage (third processing circuit), the contour data of the shape extracted in the first stage (first processing circuit) and the contour of the pattern extracted in the second stage (second processing circuit) By synthesizing the data, the contour extraction data of the object can be accurately extracted.

[0007]

Thus, according to the present invention, there is provided an image contour extraction method and apparatus for accurately extracting contour data including not only a shape but also a pattern from image data consisting of coordinate data and luminance data of an object surface. Can now be offered.

[0008]

Embodiments will be described below. The three-dimensional image input device 1 used in the present invention, for example, irradiates a measurement light beam to an object 3 arranged on a reference plane 2 on an XY plane as shown in FIG. A light source 4 composed of a laser and a light receiving element 9 composed of a CCD one-dimensional image sensor arranged along the X-axis direction are opposed to each other with a scanning mirror 5 interposed therebetween. Then, the measurement light beam from the light source 4 is main-scanned in the X-axis direction via the scanning mirror 5 and the fixed mirror 6 to irradiate the object 3 and the scattered light beam from the surface of the object 3 is fixed to the fixed mirror. The sub-scan in the Y-axis direction is performed by moving the optical head leading to the light receiving element 9 through the scanning mirror 5 and the condensing lens 8 in the Y-axis direction, as shown in FIG. (Known position) X of the scattered light beam from
₀ and the distance X ₀ X ₁ between the condensing position X ₁ of the scattered light beam from the surface of the object 3, be proportional to [Delta] X _0, and,
The distance Z ₀ from the reference plane 2 to the surface of the object 3 is Z
_{Since the} relationship of ₀ × θ = Δ _{0 is} obtained, the distance Z ₀ is obtained, and the luminance of the surface of the object 3 is obtained from the intensity detected by the light receiving element 9 at that time. The luminance data of the surface of the object 3 and the depth data represented by the z coordinate data are set for each pixel represented by the (x, y) coordinate on the XY plane to obtain image data. (X, y, z)
The coordinates indicate the shape of the surface of the object 3.

Here, in the above-described main scanning by the optical head, as shown in FIG. 5A, taking into account that the measurement light beam may be a shadow of the object 3 and cannot be measured,
As shown in (b), a scanning mechanism 10 for moving the optical head in the X-axis or Y-axis direction and moving the reference surface 2 on which the object 3 is mounted in parallel in the Y-axis or X-axis direction may be provided. . For example, the reference surface 2 may be scanned in the Y-axis direction via a rack driven by the motor 11 and the pinion mechanism 12, and the optical head may be reciprocally scanned in the X-axis direction by a similar mechanism. In this case, the scanning mirror 5 is
It is preferable to fix the angle at 5 °.

As shown in FIG. 1, the apparatus using the image contour extracting method is adapted to process image data consisting of coordinate data and luminance data of the surface of the object 3 input by the three-dimensional image input apparatus 1 described above. , The gradient of the coordinate data for each pixel is calculated and derived in 8 bits (range from 00h to FFh),
A first processing circuit 20 for extracting the contour data of the shape by binarizing the gradient with a first threshold, and calculating and deriving the gradient of the luminance data for each pixel in the same 8 bits (range from 00h to FFh) A second processing circuit 30 for masking a portion corresponding to the contour of the shape extracted in the first stage from the gradient and then binarizing the pattern with a second threshold to extract the contour data of the pattern
A third processing circuit 40 that combines the contour data of the shape extracted by the first processing circuit 20 and the contour data of the pattern extracted by the second processing circuit 30 to extract the contour data of the object 3; Consists of

As shown in FIG. 2A, a case in which a teacup having a pattern of autumn leaves drawn on its surface is used as the object 3 will be described in detail. The first processing circuit 20 converts the coordinate data (x, y, z) of the surface of the object 3 and the image data composed of the luminance data with the coordinate data ( First difference processing circuit 21 for calculating and deriving the gradient of z)
And a first binarization circuit 22 for binarizing the gradient obtained by the difference processing circuit 21 with a first threshold value and extracting the contour data of the shape as 00h and the background data as FFh. The appearance of the bowl as shown in FIG. 2 (b) is extracted as its contour data. Here, as the first threshold value, a histogram of the gradient data of the object 3 is taken in advance, and an intermediate value is set so that the gradient is gradually divided into two parts. It should be set appropriately.

The second processing circuit 30 calculates and calculates a gradient of luminance data between adjacent pixels for each pixel (x, y), and the second processing circuit 31 extracts the gradient from the gradient in the first processing circuit 20. A logic circuit 32 for masking a portion corresponding to the contour of the shaped shape (logical AND of the gradient data and the contour data of the shape extracted by the first processing circuit 20);
Of the pattern as 00h and the background data as FFh by binarizing the output data of
And a value conversion circuit 33. As for the masking by the logic circuit 32, as a portion corresponding to the contour of the shape extracted by the first processing circuit 20, in addition to masking only pixels corresponding to the contour itself, a pixel representing the contour is expanded and processed. By masking the area including the adjacent pixels, the influence from the adjacent pixels can be removed. In this way, FIG.
The outline of the pattern of the autumn leaves drawn on the surface of the teacup as shown in (1) is extracted.

The third processing circuit 40 synthesizes (constitutes a logical sum) the contour data of the shape extracted by the first processing circuit 20 and the contour data of the pattern extracted by the second processing circuit 30.
Thus, a logic circuit for extracting contour data of the object 3 is obtained, and thus a contour as shown in FIG.

Hereinafter, another embodiment of the present invention will be described. The three-dimensional image input device 1 is not limited to the above-described one, but is optional. For example, the shape data may be synthesized by synthesizing images from a plurality of directions by an imaging device such as a television camera.

Each of the first processing circuit 20, the second processing circuit 30, and the third processing circuit 40 may be constituted by hardware, or may be constituted by a computer and software for operating the computer. You may. Further, their specific configurations can be appropriately configured using a known circuit processing technique or the like.

In the above embodiment, the luminance data of the surface of the object 3 and the depth data represented by the z coordinate data are represented by X-
In the case where the image data is set for each pixel represented by the (x, y) coordinate on the Y plane and the shape data is extracted based on the value in the z-coordinate direction, the description is limited to this. Instead of extracting the shape contour data based on the value in the x-coordinate direction, for the image data set for each pixel represented by the (y, z) coordinate on the YZ plane, The contour data of the pattern may be extracted using the luminance corresponding to the (y, z) coordinates. Similarly, for each pixel represented by the (z, x) coordinates on the ZX plane, With respect to the image data to be set, the contour data of the shape is extracted based on the value in the y coordinate direction,
The contour data of the pattern may be extracted using the luminance corresponding to the (z, x) coordinates.

In the claims, reference numerals are provided for convenience of comparison with the drawings, but the present invention is not limited to the configuration shown in the attached drawings.

[Brief description of the drawings]

FIG. 1 is a block diagram of an image contour extraction device.

FIG. 2 is an explanatory diagram of an image contour extraction method.

FIG. 3 is a block diagram of a three-dimensional image input device.

FIG. 4 is an explanatory diagram showing the principle of a three-dimensional image input device.

FIG. 5 is a block diagram of a three-dimensional image input device showing another embodiment.

[Explanation of symbols]

 3 Object 20 First processing circuit 30 Second processing circuit 40 Third processing circuit

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-6-76075 (JP, A) JP-A-5-168334 (JP, A) JP-A-58-219679 (JP, A) (58) Field (Int.Cl. ⁶ , DB name) G06T 9/20

Claims (2)

(57) [Claims] 1. A gradient of coordinate data for each pixel is calculated and derived for image data composed of coordinate data and luminance data of a surface of an object input by a three-dimensional image input device, and the gradient is defined as a first gradient. A first step of extracting the contour data of the shape by binarizing it with a threshold value, and calculating and deriving a gradient of the luminance data for each pixel, and corresponding to the contour of the shape extracted in the first step from the gradient. After masking the part to be extracted, a second step of binarizing with a second threshold value to extract the contour data of the pattern, and the contour data of the shape extracted in the first step and the contour data extracted in the second step Extracting the contour data of the object by synthesizing the contour data of the pattern with the contour data of the pattern. 2. A gradient of coordinate data for each pixel is calculated and derived for image data composed of coordinate data and luminance data of the surface of an object (3) input by a three-dimensional image input device. A first processing circuit (20) for binarizing the data with a first threshold value to extract contour data of the shape, calculating and deriving a gradient of luminance data for each pixel, and extracting the gradient from the gradient in the first processing circuit. A second processing circuit (30) for extracting a contour data of the pattern by masking a portion corresponding to the contour of the formed shape and then binarizing the pattern with a second threshold value, and a shape extracted by the first processing circuit (20) And a third processing circuit (40) for extracting contour data of the object (3) by synthesizing the contour data of the target and the contour data of the pattern extracted by the second processing circuit (30). .